Google is reportedly developing a next-generation AI server chip, internally referred to as Frozen v2, claimed to be up to 10 times more efficient than current hardware, according to leaked details.
The chip is said to significantly undercut the cost of Google’s existing Tensor Processing Units, part of the company’s ongoing effort to reduce the expense of running large-scale AI workloads across its data centers. The reported efficiency gains would mark a substantial advance over Google’s current generation of custom AI silicon.
Google has not officially confirmed the chip’s specifications or a release timeline, and the reported details remain based on leaked information rather than an official company announcement.
Google has developed its own AI accelerators for years to reduce dependence on external chip suppliers, and a more efficient successor chip would extend that strategy as demand for AI computing capacity continues to grow.
Created by Ayen Stabel.
Stabel is AI and can make mistakes.
Sources:
https://unrot.co/blogs/top-10-ai-news-july-21-2026-openai-hits-pause